What Is Multiple Realizability?

Direct Answer

Multiple realizability is the thesis that a given kind of mental state can be realized, or implemented, by many different kinds of physical states. Pain, for example, can be realized in the C-fibers of a human nervous system, in the very different neural architecture of an octopus, in the exotic circuitry of an alien, or in the silicon wiring of a computer — if any of these systems can enter states that play the pain role. What makes a state a pain is not its physical constitution but its function within the organism's information-processing economy. The same mental kind, many physical kinds.

The thesis is usually contrasted with its negation: the claim, made by the type identity theory, that each mental kind is identical to one specific physical kind — that pain just is C-fiber firing, no more and no less. If multiple realizability is true, type identity is false, because identity requires that all pains share one physical nature, whereas multiple realizability says they can share none.

The concept needs three clarifications. First, "realization" is not "reduction": a realized state is implemented in a physical system, but the mental description does not reduce to the physical description, because the same mental kind can be realized in physically heterogeneous ways. Second, multiple realizability is a claim about kinds (types), not about particular events: every particular mental event must, of course, be a physical event of some kind. Third, the thesis is empirical in spirit: it is motivated by facts about biology (the diversity of nervous systems), engineering (the diversity of computing substrates), and commonsense possibility (aliens, robots, Martians with minds).

Historical Context

The argument was introduced by Hilary Putnam in his 1967 paper "Psychological Predicates" (later reprinted as "The Nature of Mental States"). Putnam was responding to the type identity theory, then the dominant physicalist position. His argument was disarmingly simple: if mental states were identical to brain states, then any creature lacking the relevant brain states would lack the mental states. But it seems obvious that creatures with very different brains — an octopus, an alien, a robot — could feel pain, hold beliefs, and have desires. Therefore mental kinds are not identical to brain kinds. They are, Putnam argued, functional kinds: kinds defined by causal role.

The argument transformed the field. Within a few years, functionalism — the view that mental states are defined by their causal roles — became the dominant framework in philosophy of mind, and nonreductive physicalism (mental properties supervene on physical properties but are not identical to them) became the default physicalist position. The argument also shaped the philosophy of science: Jerry Fodor argued in the 1970s that special sciences like psychology are autonomous because their kinds are multiply realizable and hence irreducible to the kinds of physics.

In later decades, the thesis was subjected to empirical and philosophical challenge. Lawrence Shapiro, William Bechtel, and others argued that multiple realizability is less ubiquitous than it appears and that real cognitive kinds may be realized in a narrower range of ways; Patricia Churchland argued that neuroscience, as it matures, may carve mental kinds at their neural joints. Putnam himself, in the 1990s, came to criticize the computational functionalism his argument had launched.

Key Arguments & Debates

The central argument for multiple realizability is the argument from diversity: evolution has produced wildly different nervous systems, yet psychology attributes the same mental states — pain, fear, belief, memory — across species. If pain were C-fiber firing, an octopus (which has no C-fibers) could not be in pain, which seems absurd. The argument extends to engineering: computing systems with completely different hardware run the same programs, so if minds are programs, the same mind can run on different substrates. And the modal intuition — an alien with a completely nonhuman physiology could still be happy, sad, or in pain — makes the thesis compelling.

The debate has several fronts. One is empirical: does the comparative evidence actually support the claim? Critics note that "pain" and "belief" may be multiply realizable only because they are coarse-grained; fine-grained, scientifically grounded kinds may be realized narrowly. A second front is definitional: "realization" needs a precise definition, and different definitions (realization as role-occupancy, realization as substrate, realization as implementation) yield different verdicts on which cases count. A third is its consequences: if mental kinds are multiply realizable, then psychology cannot be reduced to neuroscience, and the special sciences are autonomous — but critics like Kim argue this leads to the causal exclusion problem, in which multiply realized mental properties are causally redundant.

Defenders respond that the empirical evidence for neural diversity is real, that fine-grained kinds are not the kinds we care about when we talk about minds, and that the causal exclusion problem can be resolved by treating mental properties as realized, not additional. The debate is ongoing, and multiple realizability remains one of the most influential and contested theses in the philosophy of mind.

Contemporary Relevance

Multiple realizability is the philosophical foundation of the computational theory of mind and, through it, of artificial intelligence. The whole project of building minds in silicon presupposes that cognition can be realized in a substrate very different from the brain — exactly what multiple realizability affirms. Deep learning systems that recognize faces, translate languages, and play chess are, on the functionalist reading, realizations of cognitive capacities in new substrates.

The thesis also matters for debates about AI consciousness: if consciousness is multiply realizable, then a silicon system with the right functional organization could be conscious, regardless of substrate. And it structures the ethics of animals and machines: if mental states are multiply realizable, then the moral weight of suffering does not depend on sharing our neurobiology. Meanwhile, the empirical challenge to the thesis — the possibility that real mental kinds are realized narrowly — keeps the question live: whether a sufficiently different substrate can host a mind is, in part, an empirical question that only the future of neuroscience and AI will answer.

Related Concepts

• The identity theory: the thesis multiple realizability was designed to refute.
• Functionalism: the theory of mind that multiple realizability motivates.
• Supervenience: the dependence relation that survives multiple realizability.
• Emergentism: an alternative account of how mental properties relate to physical ones.
• What is the hard problem vs easy problems: whether consciousness itself is multiply realizable.

Further Learning

• Stanford Encyclopedia of Philosophy: Multiple Realizability
• Internet Encyclopedia of Philosophy: Multiple Realizability
• Hilary Putnam, "The Nature of Mental States," in Mind, Language and Reality (1975).
• Jerry Fodor, "Special Sciences, or the Disunity of Science as a Working Hypothesis," Synthese 28 (1974).